Re: [xacml] (local) names and issuer as string issue

[Bill Parducci <bill.parducci@overxeer.com>]

Polar Humenn wrote:
> But even then, specifying the "semantics identifier" gets you into
> interoperability problems.


'interoperability' problems or 'compatibility' problems? not to swerve 
off into pedanticLand, but i differentiate between the two of these. in 
the former you have two (or more) systems that cannot determine the 
meaning of the "semantics identifier" syntax, in the latter you have two 
(or more) systems that know what to do with the mapping itself.

the problem seems to be similar with that of issuing obligations: it is 
possible to receive back an obligation that you [PDP] don't know what to 
do with. likewise my guess is that an identify mapping that is returned 
in a format that is indecipherable is treated as an unknown subject.

the big difference here is that it seems unlikely that one would make 
such a structure optional; compliant devices will have to understand how 
to handle whatever mechanism/extension point is derived, but not 
necessarily know how to decode the responses, beyond a (single?, 
couple?, few?) normative model(s).

b